ALGORITHMIC TRADING BOT
PROJECT TEAM ID:MP23CSE141
G u i d e N a m e : D r. S u m i t P u n d i r
Associate Professor
K A RT I K AY 2016809
A M A N P R A J A PAT I 2016617
MANISH KUMAR SINGH 2016847
TA N I S H G U P TA 2017083
INTRODUCTION:
Algorithmic trading automates order execution, leveraging
computer speed and predefined rules. It enhances success rates
by refining strategies based on timing, value, and mathematical
models, bypassing human emotions for rapid, precise trades.
The trading bot simplifies processes, enabling quick profits with
minimal effort. It addresses the demand for an accessible yet
efficient solution in the expanding global algorithmic trading
market.
PROBLEM STATEMENT
Creating a dynamic algorithmic trading bot is key
in navigating modern financial markets. Focused
on agile market analysis and adaptable
strategies, it optimizes for profitability and risk
management, integrating fundamental and
sentiment analysis to seize opportunities while
minimizing risks.
LITERATURE SURVEY
Case Study 02
Advancements in AI and ML for Stock Market Prediction
• AI and ML have significantly improved stock market prediction, addressing its
complexity and dynamism.
Key Models
1. Linear and Logistic Regression Case Study 01
⚬ Linear Regression: Easy to implement but assumes linear relationships,
limiting effectiveness for complex market data.
⚬ Logistic Regression: Estimates probabilities for binary outcomes,
constrained by linearity assumptions.
LITERATURE SURVEY
Key Models Case Study 02
2. Deep Learning Models
⚬ LSTM Networks: Excellent for sequential data and long-term
dependencies, outperform traditional models but need large datasets and
are computationally expensive.
3 . Ensemble Model
Case Study 01
⚬ Random Forest: Handles large datasets, reduces overfitting but is less
interpretable and resource-demanding.
⚬ XGBoost: Efficient, scalable, performs well on structured data, requires
complex parameter tuning.
OBJECTIVES
1.Market Analysis and Strategy Development: Bot analysesmarkets,
tailoring strategies for stocks, crypto, forex, and commodities in real-
time
2.Data Acquisition and Processing: Efficiently processes vast financial
data from APIs, market feeds, and news outlets.
3.Risk Management: Implements robust risk management with stop-
loss orders, position sizing, and portfolio diversification.
4.Execution Speed and Efficiency: Prioritizes low-latency, high- trading
for swift execution and minimal slippage.
5.Back-testing and Simulation: Utilizes a concise back-testing
framework to optimize strategies and simulate diverse market
scenarios.
METHODOLOGY
• Model Selection: Based on the data analysis and
initial testing, select the most promising models
(ARIMA, LSTM, Prophet, and FinBERT) for further
development and backtesting.
• Strategy development:
⚬ ARIMA-Based Strategy: Trade based on forecasted
price trends from the ARIMA model.
⚬ LSTM-Based Strategy: Use LSTM predictions to
make informed trading decisions.
METHODOLOGY
• Strategy development:
⚬ Sentiment-Based Strategy with FinBERT: React to
market sentiment using FinBERT to trigger buy or
sell signals.
⚬ ROC Strategy:The trading bot calculates the rate of
change (ROC) for stocks, identifies buy signals by
comparing ask prices with the last traded prices,
and sends an email alert for the stock with the
highest ROC and favorable price conditions.
METHODOLOGY
• Signal Generation: Generate trading signals based on
model predictions; e.g., a predicted upward trend triggers
a buy signal, a downward trend triggers a sell signal.
• Risk Management: Incorporate risk management
techniques like stop-loss levels, take-profit targets, and
position sizing to manage losses and protect profits.
• Portfolio Allocation: Allocate capital across different
strategies, using a diversified approach based on each
strategy's expected performance and risk profile.
• Continuous Monitoring: Set up a system to track each
strategy's performance in real-time, including monitoring
trading signals, executed trades, and overall portfolio
performance.
METHODOLOGY
• Back Testing:
⚬ Data Collection: Acquire comprehensive historical
market data, ensuring cleanliness and accuracy.
⚬ Strategy Definition: Clearly define trading rules
and specify parameters like moving averages and
stop-loss levels.
⚬ Simulation: Execute trades on historical data
according to predefined rules, simulating order
handling and potential slippage.
METHODOLOGY
• Back Testing:
⚬ Performance Metrics Calculation: Calculate total
and annualized returns, evaluate risk metrics, and
analyze trade statistics.
⚬ Analysis and Optimization: Analyze results,
optimize parameters cautiously, and validate
robustness using out-of-sample testing.
RESULTS AND DISCUSSION
THE CHART
ALONGSIDE
REPRESENTS THE
GAINS OBTAINED
BY USING THE ROC
AND NLP BASED
STRATEGIES FOR
TRADING
ONE MONTH GAINS ARE 8.345%
KEY PERFORMANCE
INDICATORS
THESE RESULTS ARE PROOF THAT STRATEGY IS
QUITE ROBUST AS IT WAS ABLE TO MAKE
SUBSTANTIAL PROFITS IN 2022 WHEN MARKETS
WERE VOLATILE AND ECONOMIES WERE
RECOVERING FROM THE PANDEMIC
RETURNS
COMPARED
TO S&P 500
The graph alongside
represents the cumulative
return as compared to the
SPDR S&P 500 ETF trust
fund. SPDR is an acronym
for the Standard & Poor's
Depositary Receipts, the
former name of the ETF. It is
designed to track the S&P
500 stock market index. This GRAPH BASED ON 5 YEARS OF
fund is the largest and oldest BACKTESTING
ETF in the USA.
Monthly Returns Barplot
The barplot represents the monthly returns of
the trading strategy. The returns kind of follow a
bell shaped curve. The gains have a wide spread
resulting which signifies high volatility of the
NLP strategy with 100 percent cash at risk.
Heatmap
As evident from the heatmap, we can see that
strategy is very volatile if 100 % of cash is at
risk. Though this ultimately gives high return in
future, but such risky practices should be
avoided and traders should try to square off at
their threshold gains
PERFORMANCE OF IMPLEMENTED STRATEGY COMPARED TO
S&P 500 ETF TRUST FUND
CONCLUSION AND FUTURE
WORK
The project successfully developed an advanced algorithmic trading bot
using ARIMA, LSTM, and FinBERT models, validated through extensive
backtesting and live trading simulations, demonstrating consistent
returns and effective risk management. Future improvements include
implementing high-frequency trading algorithms to enhance execution
speed and precision, and developing a user-friendly dashboard for real-
time insights and analytics, aiming to create a robust, efficient, and
user-centric trading solution.
REFERENCES
[1] Ash Booth, Enrico Gerding, and Frank Mcgroarty. 2014. Automated trading with performance weighted random forests
and seasonality. Expert Syst. Appl. 41, 8 (June ,2014), 3651± 3661. DOI: https://doi.org/10.1016/j.eswa.2013.12.009.
[2] Suryoday Basak, Saibal Kar, Snehanshu Saha, Luckyson Khaidem, Sudeepa Roy Dey, Predicting the direction of stock
market prices using tree-based classifiers, The North American Journal of Economics and Finance, Volume 47, 2019, Pages
552- 567, ISSN 1062-9408,
https://doi.org/10.1016/j.najef.2018.06.013.
[3] Maragoudakis, M and Serpanos, D. (2010), towards stock market data mining using enriched random forests from
textual resources and technical indicators. AIAI 2010, IFIP AICT 339, pp. 278-286.
https://link.springer.com/chapter/10.1007/978-3-642-16239-8_37.
[4]Manoj Thakur and Deepak Kumar. 2018. A hybrid financial trading support system using multi-category classifiers and
random forest. Appl. Soft Compute. 67, C (June 2018), 337±349. DOI: https://doi.org/10.1016/j.asoc.2018.03.006.
[5] Cain Evans, Konstantinos Pappas, Fatos Xhafa, Utilizing artificial neural networks and
genetic algorithms to build an algo-trading model for intra-day foreign exchange speculation, Mathematical and
Computer Modelling, Volume 58, Issues 5±6, 2013, Pages 1249-1266, ISSN 0895-7177,
https://doi.org/10.1016/j.mcm.2013.02.002.
[6] Omer Berat Sezer, Murat Ozbayoglu, Erdogan Dogdu, A Deep Neural-Network Based Stock Trading System Based
on Evolutionary Optimized Technical Analysis Parameters, Procedia Computer Science, Volume 114, 2017, Pages 473-
480, ISSN 1877-0509
[7]Maragoudakis, M and Serpanos, D. (2010), towards stock market data mining using enriched random forests from
textual resources and technical indicators. AIAI 2010, IFIP AICT 339, pp. 278-286.
[8] Manoj Thakur and Deepak Kumar. 2018. A hybrid financial trading support system using multi-category classifiers
and random forest. Appl. Soft Comput. 67, C (June 2018), 337±349. DOI:
THANK YOU
